The Philadelphia Fed's Forecaster Panel Lowered Its Recession Odds in Every Quarter Both Surveys Covered. It Also Cut This Quarter's Job-Gain Pace.

by Tomas Ferreira · August 17, 2026

Markets Reporter, US Market Current

The Federal Reserve Bank of Philadelphia released its Third Quarter 2026 Survey of Professional Forecasters on Friday, Aug. 14. It is the oldest quarterly survey of macroeconomic forecasts in the United States, and this edition drew 32 respondents. The previous edition, released Friday, May 15, drew 33.

The release opens by telling readers where it lands: "The near-term outlook for the U.S. economy looks more positive now than it did three months ago, according to 32 forecasters surveyed by the Federal Reserve Bank of Philadelphia." The tables behind that sentence are consistent with it on growth, on unemployment and on the panel's stated odds of a contraction. They are not uniformly consistent with it on the labor market, where the estimate of job gains in the current quarter was cut. On inflation the answer depends on the horizon: the 2026 fourth-quarter-over-fourth-quarter figures moved by a tenth or not at all, while the estimate for the quarter now under way was marked down by seven-tenths of a percentage point.

This article covers one document — the Philadelphia Fed's third-quarter survey — and its line-by-line comparison with the second-quarter survey that the release itself prints alongside it. It does not attempt to reconcile the survey against any other forecast, and it does not cite any index level.

Start with the number the Philadelphia Fed publishes under its own name. The release includes a table titled "Risk of a Negative Quarter (%) Survey Means." Its leftmost column carries the quarter labels; the two value columns to the right of it are headed "Previous" and "New." Read against those quarter labels, the New column shows a mean probability of a decline in real GDP of 13.3 percent for 2026:Q3, the current quarter, 20.0 percent for the fourth quarter of 2026, 21.5 percent for the first quarter of 2027, 22.0 percent for the second quarter of 2027 and 20.8 percent for the third quarter of 2027.

The Previous column reproduces the May survey's readings against those same quarter labels: 25.1 for 2026:Q3, 24.5 for 2026:Q4, 25.7 for 2027:Q1 and 23.0 for 2027:Q2. The 2027:Q3 row is marked N.A. because the May panel's table did not reach that far. Comparing the two columns row by row — an aggregation this publication performed from the printed table, not a summary the Philadelphia Fed states in its own prose — every one of the four quarters the two surveys have in common came down. The largest single move is in the current quarter, 25.1 to 13.3, a decline of 11.8 percentage points, and 13.3 percent is the lowest of the five probabilities printed in the New column.

The Philadelphia Fed does state that current-quarter move in its own words: "The forecasters have reduced their estimate of the risk of a downturn in real GDP this quarter to 13.3 percent, compared with the estimate of 25.1 percent." The second of the five figures has a separate public life. The Bank maintains a standing page for what it calls the anxious index, which it defines as the probability of a decline in real GDP in the quarter after a survey is taken. For a third-quarter survey that is the fourth quarter, and the page — last updated Aug. 14 — carries the same 20.0 percent for 2026:Q4. That page also sets out what the series has done historically: "The index often goes up just before recessions begin." The example it gives is the first-quarter survey of 2001, taken in February, which "reported a 32 percent anxious index; the National Bureau of Economic Research subsequently declared the start of a recession in March 2001." The Bank adds that the index "peaks during recessions, then declines when recovery seems near." Friday's 20.0 percent is well below that 2001 reading.

The growth and labor market table is titled "Median Forecasts for Selected Variables." It carries three variable groups — Real GDP (%), Unemployment Rate (%) and Payrolls (000s/month) — and under each group two value columns, the left one reproducing the May survey and the right one the August survey, labelled Previous and New. On real GDP the release puts the panel at an annual rate of 2.5 percent this quarter and 2.3 percent next quarter, "up from the predictions of 2.2 percent and 1.6 percent in the last survey."

All four quarterly real GDP rows the two surveys share moved up in the New column — 2.2 to 2.5 for 2026:Q3, 1.6 to 2.3 for 2026:Q4, and 1.9 to 2.1 for each of the first two quarters of 2027. The annual rows did not all move the same way. The median for calendar 2026 slipped a tenth, from 2.2 percent to 2.1 percent, and the 2028 median slipped a tenth as well, from 2.2 percent to 2.1 percent. The 2027 median rose from 1.9 percent to 2.2 percent and the 2029 median rose from 2.3 percent to 2.4 percent.

Unemployment moved one way only. The release says the panel expects a rate that "will range from 4.2 percent to 4.3 percent" over the stretch running from this quarter through the second quarter of 2027. In the Previous column every one of those four quarterly rows read 4.5 percent. On an annual basis the New column shows 4.3 percent for 2026, 4.3 percent for 2027, 4.2 percent for 2028 and 4.3 percent for 2029, against 4.4, 4.5, 4.4 and 4.4 in the Previous column.

The payroll column is where the direction changes. The release reports job gains in the current quarter at a rate of 45,600 per month, against a previous estimate of 61,200. That is a reduction of 15,600 jobs a month in the quarter the panel is living through, printed in the same table and the same survey that raised the growth path and cut the contraction odds. The column is labelled in thousands per month, so the two figures appear on the page as 45.6 and 61.2.

The rest of the payroll column is mixed rather than weak. The 2026:Q4 row rose from 58.4 to 66.4, and the 2027:Q1 row rose from 64.0 to 94.9 — the largest quarterly increase anywhere in that column. The 2027:Q2 row fell from 97.3 to 80.4. Both annual rows went up: the 2026 average rose from 34.6 to 40.7 and the 2027 average from 64.4 to 75.4. So of the four overlapping quarterly rows, two were marked down and two were marked up, while the two annual figures were both marked up.

Inflation is the part of this release where the horizon decides the story. The relevant table is titled "Median Short-Run and Long-Run Projections for Inflation (Annualized Percentage Points)," and its sub-columns are labelled "Previous" and "Current" rather than Previous and New. In the section headed "Q4/Q4 Annual Averages," the 2026 row shows headline CPI moving from 3.5 percent to 3.6 percent and core CPI moving from 2.9 percent to 2.8 percent. Headline PCE stayed at 3.6 percent and core PCE at 3.3 percent.

Those are moves of a tenth in opposite directions on the CPI measures and no move at all on the two PCE measures. The shorter horizon is a different matter. In the quarterly rows of the same table, the panel cut its 2026:Q3 headline CPI projection from 3.0 percent to 2.3 percent and its core CPI projection from 2.9 percent to 2.7 percent, and cut headline PCE for the same quarter from 3.0 percent to 2.3 percent. The 2027 Q4/Q4 headline CPI figure came down from 2.5 percent to 2.3 percent. The panel did not leave inflation untouched: it left its 2026 fourth-quarter-over-fourth-quarter view roughly where it was while marking down the quarter now under way by seven-tenths of a percentage point.

Over the longer horizon the table's 10-year row, covering 2026 to 2035, puts headline CPI at an annual-average rate of 2.30 percent, a tenth below the 2.40 percent printed in the Previous column. The five-year row, covering 2026 to 2030, is unchanged at 2.60 percent.

Because a two-column table is exactly where a reading can go wrong, the Previous column was checked against its own source rather than assumed. The May 15 survey's own current-column values for 2026:Q3 and 2026:Q4 are 2.2 and 1.6 on real GDP, 4.5 and 4.5 on unemployment, and 61.2 and 58.4 on payrolls; its risk-of-a-negative-quarter table runs from 2026:Q2 through 2027:Q2 and reads 17.9, 25.1, 24.5, 25.7 and 23.0; and its 2026 Q4/Q4 headline and core CPI medians are 3.5 percent and 2.9 percent. Every figure that reappears in Friday's Previous column matches the May survey quarter label for quarter label. The May table's leading row, 2026:Q2, has no counterpart in the August survey, which is why the two surveys overlap in four quarters rather than five.

Three limits are worth stating plainly. The survey reports medians for the forecast variables and means for the probability questions — they describe a panel of private-sector forecasters, not the Federal Reserve's own outlook, and not the projections that FOMC participants submit. The probability figures are estimates of the chance of a decline in real GDP in a given quarter, which is not the same thing as the National Bureau of Economic Research's dating of a recession. And a panel lowering its stated odds of a contraction is a change in what 32 forecasters expect, not a change in the economy itself; none of these figures is a measurement of output, employment or prices already recorded.